Case Studies - Lecture 12 - Information Visualisation (4019538FNR)

This lecture forms part of the course Information Visualisation given at the Vrije Universiteit Brussel.

Transcript

1. 2 December 2005 Information Visualisation Case Studies Prof. Beat Signer Department of Computer Science Vrije Universiteit Brussel beatsigner.com Department of Computer Science Vrije Universiteit Brussel beatsigner.com
2. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 2 May 13, 2026 Analyse Case Studies ▪ Analysis of existing systems provides foundation for considering all the possibilities when designing new systems ▪ use analysis framework introduced earlier - what, why and how? - four levels of validation ▪ data/task abstraction - types of data abstraction - derived data - … ▪ visual encoding/interaction idioms - encoding design choices - faceting between multiple views - …
3. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 3 May 13, 2026 Scagnostics SPLOM ▪ Scalable idiom for the exploration of scatterplot matrices (SPLOMs) ▪ scagnostics = scatterplot computer-guided diagnostics
4. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 4 May 13, 2026 Scagnostics SPLOM … ▪ Use nine measurements that categorise the point distribution of scatterplots ▪ monotonic, stringy, skinny, convex, striated, sparse, clumpy, skewed and outlying ▪ Show measurements in a new scagnostics SPLOM ▪ scatterplot of scatterplots ▪ each point in the scagnostics SPLOM represents an entire scatterplot of the original SPLOM
5. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 5 May 13, 2026 Scagnostics SPLOM …
6. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 6 May 13, 2026 Scagnostics SPLOM … ▪ Linked highlighting between views ▪ Selection of point triggers popup view with full scatterplot Scagnostics SPLOM What (Data) Table. What (Derived) Nine quantitative attributes per scatterplot (pairwise combination of original attributes). Why (Tasks) Identify, compare, and summarise; distributions and correlation. How (Encode) Scatterplot, scatterplot matrix. How (Manipulate) Select. How (Facet) Juxtaposed small-multiple views coordinated with linked highlighting, popup detail view. Scale Original attributes: dozens.
7. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 7 May 13, 2026 Hierarchical Clustering Explorer (HCE) ▪ Systematic exploration of multidimensional table ▪ Originally designed for genomics domain ▪ multidimensional table with two key attributes (genes and experimental conditions) and a quantitative value attribute (activity of gene under experimental condition) ▪ derived data is a cluster hierarchy of items based on a similarity measure between items ▪ scalability target: 100-20'000 gene attributes and 2-80 experimental condition attributes ▪ Scalability through combination of visual encoding and interaction idioms
8. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 8 May 13, 2026 Hierarchical Clustering Explorer (HCE) …
9. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 9 May 13, 2026 Hierarchical Clustering Explorer (HCE) … Hierarchical Clustering Explorer (HCE) What (Derived) Hierarchical clustering of table rows and columns (for cluster heatmap); quantitative derived attributes for each attribute and pairwise attribute combination; quantitative derived attribute for each ranking criterion and original attribute combination. Why (Tasks) Find correlation between attributes; find clusters, gaps, outliers, trends within items. How (Encode) Cluster heatmap, scatterplots, histograms. How (Reduce) Dynamic filtering; dynamic aggregation. How (Manipulate) Navigate with pan/scroll. How (Facet) Multiform with linked highlighting and shared spatial position; overview-detail with selection in overview populating detail view Scale Genes (key attribute): 20'000. Conditions (key attribute): 80. Gene activity in condition (quantitative value attribute): 20'000 × 80 = 1'600'000.
10. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 10 May 13, 2026 PivotGraph ▪ PivotGraph idiom encodes a network derived from the original network by aggregating groups of nodes and links into a roll-up ▪ grouping based on categorical attribute values on the nodes (up to two attributes)
11. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 11 May 13, 2026 PivotGraph …
12. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 12 May 13, 2026 PivotGraph … ▪ PivotGraph idiom is highly scalable ▪ summarises arbitrarily large number of nodes and links of the original network ▪ Visual complexity of the derived network depends on the number of attribute levels for the two roll-up attributes ▪ PivotGraph complements standard encoding idioms for networks (e.g.node-link and matrix views) ▪ might be used as a linked multiform view ▪ Well suited for comparison across attributes at the aggregate level ▪ but not good to understand topological network features
13. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 13 May 13, 2026 PivotGraph … PivotGraph What (Data) Network. What (Derived) Derived network of aggregate nodes and links by roll-up into two chosen attributes. Why (Task) Cross-attribute comparison of node groups. How (Encode) Nodes linked with connection marks, size. How (Manipulate) Change: animated transitions. How (Reduce) Aggregation, filtering. Scale Nodes/links in original network: unlimited. Rollup attributes: 2. Levels per roll-up attribute: several, up to one dozen.
14. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 14 May 13, 2026 InterRing ▪ Visual encoding and interaction idioms for tree exploration ▪ space-filling radial layout for encoding the hierarchy ▪ multifocus focus+context distortion approach for interaction ▪ structure-based colouring (redundant) - useful if shared colour coding used to coordinate with other views original hierarchy selected blue region enlarged selected tan region enlarged
15. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 15 May 13, 2026 InterRing … ▪ Works well in combination with other views ▪ hierarchy view supports selection, navigation and roll-up/drill- down operations ▪ supports direct editing of the hierarchy InterRing What (Data) Tree. Why (Task) Selection, rollup/drilldown, hierarchy editing. How (Encode) Radial, space-filling layout. Colour by tree structure. How (Facet) Linked colouring and highlighting. How (Reduce) Embed: distort; multiple foci. Scale Nodes: hundreds if labelled, thousands if dense. Levels in tree: dozens.
16. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 16 May 13, 2026 Video: The Beauty of Data Visualisation
17. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 17 May 13, 2026 Video: The Simple Genius of a Good Graphic
18. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 18 May 13, 2026 Video: Mapping Ideas Worth Spreading
19. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 19 May 13, 2026 Further Reading ▪ This lecture is mainly based on the book Visualization Analysis & Design ▪ chapter 15 - Analysis Case Studies
20. Beat Signer - Department of Computer Science - bsigner@vub.ac.be 20 May 13, 2026 References ▪ Visualization Analysis & Design, Tamara Munzner, Taylor & Francis Inc, (Har/Psc edition), May, November 2014, ISBN-13: 978-1466508910 ▪ The Beauty of Data Visualization ▪ https://www.youtube.com/watch?v=5Zg-C8AAIGg ▪ The Simple Genius of a Good Graphic ▪ https://www.youtube.com/watch?v=6C_-VdaXgCQ ▪ Mapping Ideas Worth Spreading ▪ https://www.youtube.com/watch?v=kv_uyUTx5Po
21. 2 December 2005 Next Lecture Course Review